skema-core 2.1.1 → 2.1.2

package/dist/server.d.mts

@@ -451,14 +451,19 @@ declare function isProviderAvailable(provider: AIProvider): boolean;
  */
 declare function getAvailableProviders(): AIProvider[];
 
+type VisionProvider = 'gemini' | 'claude' | 'openai';
+declare const VISION_MODELS: Record<VisionProvider, {
+    models: string[];
+    default: string;
+}>;
 interface VisionAnalysisResult {
     success: boolean;
     description: string;
-    provider: AIProvider;
+    provider: VisionProvider;
     error?: string;
 }
 interface VisionConfig {
-    provider: AIProvider;
+    provider: VisionProvider;
     /** API key for vision API (falls back to env vars) */
     apiKey?: string;
     /** Model to use for vision */
@@ -471,7 +476,7 @@ declare function analyzeImage(base64Image: string, config: VisionConfig): Promis
 /**
  * Check if vision analysis is available for a provider
  */
-declare function isVisionAvailable(provider: AIProvider): boolean;
+declare function isVisionAvailable(provider: VisionProvider): boolean;
 
 type AnnotationStatus = 'pending' | 'acknowledged' | 'resolved' | 'dismissed';
 interface StoredAnnotation {
@@ -538,4 +543,4 @@ declare function onStoreEvent(listener: StoreListener): () => void;
  */
 declare function getPendingCount(): number;
 
-export { type AIProviderConfig, type AIRunResult, type AIStreamEvent, type AnnotationStatus, type AIProvider as CLIProvider, DELETE, type DaemonConfig, type DaemonInstance, type DetailedDomSelectionInput, type DomSelectionInput, type DrawingInput, type ExecutionMode, type GeminiCLIEvent, type GeminiCLIOptions, type GestureInput, IMAGE_ANALYSIS_PROMPT, type IncomingMessage, type OutgoingMessage, POST, type ProjectContext, type ProviderName, type StoredAnnotation, type VisionAnalysisResult, type VisionConfig, acknowledgeAnnotation, analyzeImage, buildDetailedDomSelectionPrompt, buildDrawingToCodePrompt, buildFastDomSelectionPrompt, buildGesturePrompt, buildPromptFromAnnotation, clearAnnotations, createGeminiCLIStream, createGeminiRouteHandler, createRevertRouteHandler, dismissAnnotation, getAllAnnotations, getAnnotation, getAvailableProviders as getCLIProviders, getPendingAnnotations, getPendingCount, getTrackedAnnotations, isProviderAvailable, isVisionAvailable, onStoreEvent, queueAnnotation, removeAnnotation, resolveAnnotation, revertAnnotation, runAICLI, runGeminiCLI, spawnAICLI, spawnGeminiCLI, startDaemon };
+export { type AIProviderConfig, type AIRunResult, type AIStreamEvent, type AnnotationStatus, type AIProvider as CLIProvider, DELETE, type DaemonConfig, type DaemonInstance, type DetailedDomSelectionInput, type DomSelectionInput, type DrawingInput, type ExecutionMode, type GeminiCLIEvent, type GeminiCLIOptions, type GestureInput, IMAGE_ANALYSIS_PROMPT, type IncomingMessage, type OutgoingMessage, POST, type ProjectContext, type ProviderName, type StoredAnnotation, VISION_MODELS, type VisionAnalysisResult, type VisionConfig, type VisionProvider, acknowledgeAnnotation, analyzeImage, buildDetailedDomSelectionPrompt, buildDrawingToCodePrompt, buildFastDomSelectionPrompt, buildGesturePrompt, buildPromptFromAnnotation, clearAnnotations, createGeminiCLIStream, createGeminiRouteHandler, createRevertRouteHandler, dismissAnnotation, getAllAnnotations, getAnnotation, getAvailableProviders as getCLIProviders, getPendingAnnotations, getPendingCount, getTrackedAnnotations, isProviderAvailable, isVisionAvailable, onStoreEvent, queueAnnotation, removeAnnotation, resolveAnnotation, revertAnnotation, runAICLI, runGeminiCLI, spawnAICLI, spawnGeminiCLI, startDaemon };

package/dist/server.d.ts

@@ -451,14 +451,19 @@ declare function isProviderAvailable(provider: AIProvider): boolean;
  */
 declare function getAvailableProviders(): AIProvider[];
 
+type VisionProvider = 'gemini' | 'claude' | 'openai';
+declare const VISION_MODELS: Record<VisionProvider, {
+    models: string[];
+    default: string;
+}>;
 interface VisionAnalysisResult {
     success: boolean;
     description: string;
-    provider: AIProvider;
+    provider: VisionProvider;
     error?: string;
 }
 interface VisionConfig {
-    provider: AIProvider;
+    provider: VisionProvider;
     /** API key for vision API (falls back to env vars) */
     apiKey?: string;
     /** Model to use for vision */
@@ -471,7 +476,7 @@ declare function analyzeImage(base64Image: string, config: VisionConfig): Promis
 /**
  * Check if vision analysis is available for a provider
  */
-declare function isVisionAvailable(provider: AIProvider): boolean;
+declare function isVisionAvailable(provider: VisionProvider): boolean;
 
 type AnnotationStatus = 'pending' | 'acknowledged' | 'resolved' | 'dismissed';
 interface StoredAnnotation {
@@ -538,4 +543,4 @@ declare function onStoreEvent(listener: StoreListener): () => void;
  */
 declare function getPendingCount(): number;
 
-export { type AIProviderConfig, type AIRunResult, type AIStreamEvent, type AnnotationStatus, type AIProvider as CLIProvider, DELETE, type DaemonConfig, type DaemonInstance, type DetailedDomSelectionInput, type DomSelectionInput, type DrawingInput, type ExecutionMode, type GeminiCLIEvent, type GeminiCLIOptions, type GestureInput, IMAGE_ANALYSIS_PROMPT, type IncomingMessage, type OutgoingMessage, POST, type ProjectContext, type ProviderName, type StoredAnnotation, type VisionAnalysisResult, type VisionConfig, acknowledgeAnnotation, analyzeImage, buildDetailedDomSelectionPrompt, buildDrawingToCodePrompt, buildFastDomSelectionPrompt, buildGesturePrompt, buildPromptFromAnnotation, clearAnnotations, createGeminiCLIStream, createGeminiRouteHandler, createRevertRouteHandler, dismissAnnotation, getAllAnnotations, getAnnotation, getAvailableProviders as getCLIProviders, getPendingAnnotations, getPendingCount, getTrackedAnnotations, isProviderAvailable, isVisionAvailable, onStoreEvent, queueAnnotation, removeAnnotation, resolveAnnotation, revertAnnotation, runAICLI, runGeminiCLI, spawnAICLI, spawnGeminiCLI, startDaemon };
+export { type AIProviderConfig, type AIRunResult, type AIStreamEvent, type AnnotationStatus, type AIProvider as CLIProvider, DELETE, type DaemonConfig, type DaemonInstance, type DetailedDomSelectionInput, type DomSelectionInput, type DrawingInput, type ExecutionMode, type GeminiCLIEvent, type GeminiCLIOptions, type GestureInput, IMAGE_ANALYSIS_PROMPT, type IncomingMessage, type OutgoingMessage, POST, type ProjectContext, type ProviderName, type StoredAnnotation, VISION_MODELS, type VisionAnalysisResult, type VisionConfig, type VisionProvider, acknowledgeAnnotation, analyzeImage, buildDetailedDomSelectionPrompt, buildDrawingToCodePrompt, buildFastDomSelectionPrompt, buildGesturePrompt, buildPromptFromAnnotation, clearAnnotations, createGeminiCLIStream, createGeminiRouteHandler, createRevertRouteHandler, dismissAnnotation, getAllAnnotations, getAnnotation, getAvailableProviders as getCLIProviders, getPendingAnnotations, getPendingCount, getTrackedAnnotations, isProviderAvailable, isVisionAvailable, onStoreEvent, queueAnnotation, removeAnnotation, resolveAnnotation, revertAnnotation, runAICLI, runGeminiCLI, spawnAICLI, spawnGeminiCLI, startDaemon };

package/dist/server.js

@@ -5,6 +5,10 @@ var generativeAi = require('@google/generative-ai');
 var ws = require('ws');
 var fs = require('fs');
 var path = require('path');
+var ai = require('ai');
+var google = require('@ai-sdk/google');
+var anthropic = require('@ai-sdk/anthropic');
+var openai = require('@ai-sdk/openai');
 
 function _interopNamespace(e) {
   if (e && e.__esModule) return e;
@@ -903,57 +907,83 @@ function getAllProviderStatuses() {
     claude: getProviderStatus("claude")
   };
 }
-async function analyzeWithGemini(base64Image, apiKey, model = "gemini-2.5-flash") {
-  try {
-    const genAI = new generativeAi.GoogleGenerativeAI(apiKey);
-    const visionModel = genAI.getGenerativeModel({ model });
-    const cleanBase64 = base64Image.replace(/^data:image\/\w+;base64,/, "");
-    const result = await visionModel.generateContent([
-      IMAGE_ANALYSIS_PROMPT,
-      {
-        inlineData: {
-          data: cleanBase64,
-          mimeType: "image/png"
-        }
-      }
-    ]);
-    const response = await result.response;
-    const text = response.text();
-    return {
-      success: true,
-      description: text,
-      provider: "gemini"
-    };
-  } catch (error) {
-    const message = error instanceof Error ? error.message : String(error);
-    console.error("[Vision] Gemini analysis failed:", message);
+var VISION_MODELS = {
+  gemini: {
+    models: ["gemini-2.5-flash", "gemini-2.5-pro", "gemini-3-flash-preview", "gemini-3-pro-preview"],
+    default: "gemini-2.5-flash"
+  },
+  claude: {
+    models: ["claude-haiku-4-5-20251001", "claude-sonnet-4-5-20250929", "claude-opus-4-6"],
+    default: "claude-haiku-4-5-20251001"
+  },
+  openai: {
+    models: ["gpt-4o-mini", "gpt-4o", "gpt-4.1", "gpt-5.2"],
+    default: "gpt-4o-mini"
+  }
+};
+function getProviderModel(provider, apiKey, model) {
+  const modelId = model || VISION_MODELS[provider].default;
+  switch (provider) {
+    case "gemini": {
+      const google$1 = google.createGoogleGenerativeAI({ apiKey });
+      return google$1(modelId);
+    }
+    case "claude": {
+      const anthropic$1 = anthropic.createAnthropic({ apiKey });
+      return anthropic$1(modelId);
+    }
+    case "openai": {
+      const openai$1 = openai.createOpenAI({ apiKey });
+      return openai$1(modelId);
+    }
+  }
+}
+function getEnvVarForProvider(provider) {
+  switch (provider) {
+    case "gemini":
+      return process.env.GEMINI_API_KEY;
+    case "claude":
+      return process.env.ANTHROPIC_API_KEY;
+    case "openai":
+      return process.env.OPENAI_API_KEY;
+  }
+}
+function getEnvVarName(provider) {
+  switch (provider) {
+    case "gemini":
+      return "GEMINI_API_KEY";
+    case "claude":
+      return "ANTHROPIC_API_KEY";
+    case "openai":
+      return "OPENAI_API_KEY";
+  }
+}
+async function analyzeImage(base64Image, config) {
+  const { provider } = config;
+  const apiKey = config.apiKey || getEnvVarForProvider(provider);
+  if (!apiKey) {
     return {
       success: false,
       description: "",
-      provider: "gemini",
-      error: message
+      provider,
+      error: `No API key found for ${provider} vision. Set ${getEnvVarName(provider)} environment variable.`
     };
   }
-}
-async function analyzeWithClaude(base64Image, apiKey, model = "claude-sonnet-4-20250514") {
+  console.log(`[Vision] Analyzing image with ${provider}...`);
   try {
-    const Anthropic = (await import('@anthropic-ai/sdk')).default;
-    const client = new Anthropic({ apiKey });
     const cleanBase64 = base64Image.replace(/^data:image\/\w+;base64,/, "");
-    const response = await client.messages.create({
+    const model = getProviderModel(provider, apiKey, config.model);
+    const result = await ai.generateText({
       model,
-      max_tokens: 1024,
+      maxTokens: 1024,
       messages: [
         {
           role: "user",
           content: [
             {
               type: "image",
-              source: {
-                type: "base64",
-                media_type: "image/png",
-                data: cleanBase64
-              }
+              image: Buffer.from(cleanBase64, "base64"),
+              mimeType: "image/png"
             },
             {
               type: "text",
@@ -963,51 +993,24 @@ async function analyzeWithClaude(base64Image, apiKey, model = "claude-sonnet-4-2
         }
       ]
     });
-    const textContent = response.content.find((c) => c.type === "text");
-    const description = textContent && "text" in textContent ? textContent.text : "";
     return {
       success: true,
-      description,
-      provider: "claude"
+      description: result.text,
+      provider
     };
   } catch (error) {
     const message = error instanceof Error ? error.message : String(error);
-    console.error("[Vision] Claude analysis failed:", message);
-    return {
-      success: false,
-      description: "",
-      provider: "claude",
-      error: message
-    };
-  }
-}
-async function analyzeImage(base64Image, config) {
-  const { provider } = config;
-  let apiKey = config.apiKey;
-  if (!apiKey) {
-    apiKey = provider === "gemini" ? process.env.GEMINI_API_KEY : process.env.ANTHROPIC_API_KEY;
-  }
-  if (!apiKey) {
+    console.error(`[Vision] ${provider} analysis failed:`, message);
     return {
       success: false,
       description: "",
       provider,
-      error: `No API key found for ${provider} vision. Set ${provider === "gemini" ? "GEMINI_API_KEY" : "ANTHROPIC_API_KEY"} environment variable.`
+      error: message
     };
   }
-  console.log(`[Vision] Analyzing image with ${provider}...`);
-  if (provider === "gemini") {
-    return analyzeWithGemini(base64Image, apiKey, config.model);
-  } else {
-    return analyzeWithClaude(base64Image, apiKey, config.model);
-  }
 }
 function isVisionAvailable(provider) {
-  if (provider === "gemini") {
-    return !!process.env.GEMINI_API_KEY;
-  } else {
-    return !!process.env.ANTHROPIC_API_KEY;
-  }
+  return !!getEnvVarForProvider(provider);
 }
 
 // src/server/annotation-store.ts
@@ -1251,9 +1254,14 @@ var handlers = {
           provider: requestProvider
         }
       });
-      if (isVisionAvailable("gemini")) {
+      const visionProvider = msg.visionProvider || "gemini";
+      const visionModel = msg.visionModel;
+      const visionApiKey = msg.visionApiKey || (visionProvider === "gemini" ? process.env.GEMINI_API_KEY : visionProvider === "claude" ? process.env.ANTHROPIC_API_KEY : process.env.OPENAI_API_KEY);
+      if (visionApiKey) {
         const visionResult = await analyzeImage(drawingAnnotation.drawingImage, {
-          provider: "gemini"
+          provider: visionProvider,
+          apiKey: visionApiKey,
+          model: visionModel
         });
         if (visionResult.success) {
           visionDescription = visionResult.description;
@@ -1286,7 +1294,7 @@ ${visionDescription}`,
           type: "ai-event",
           event: {
             type: "text",
-            content: `[Vision not available - set GEMINI_API_KEY for image analysis]`,
+            content: `[Vision not available - add your API key in Settings (gear icon)]`,
             timestamp: (/* @__PURE__ */ new Date()).toISOString(),
             provider: requestProvider
           }
@@ -1297,6 +1305,8 @@ ${visionDescription}`,
       fastMode: msg.fastMode === true,
       visionDescription
     });
+    console.log(`[Skema] Prompt:
+${prompt}`);
     sendMessage(ws, {
       id: msg.id,
       type: "ai-event",
@@ -1327,7 +1337,14 @@ ${visionDescription}`,
       model: msg.model
     };
     const { process: aiProcess, events } = spawnAICLI(prompt, config);
+    const prefixGreen = "\x1B[32m";
+    const reset = "\x1B[0m";
     for await (const event of events) {
+      if (event.type === "text" && event.content) {
+        console.log(`${prefixGreen}[Skema ${requestProvider}]${reset} ${event.content}`);
+      } else if (event.type === "error" && event.content) {
+        console.error(`${prefixGreen}[Skema ${requestProvider}]${reset} ${event.content}`);
+      }
       sendMessage(ws, {
         id: msg.id,
         type: "ai-event",
@@ -1720,6 +1737,7 @@ function startDaemon(config = {}) {
 exports.DELETE = DELETE;
 exports.IMAGE_ANALYSIS_PROMPT = IMAGE_ANALYSIS_PROMPT;
 exports.POST = POST;
+exports.VISION_MODELS = VISION_MODELS;
 exports.acknowledgeAnnotation = acknowledgeAnnotation;
 exports.analyzeImage = analyzeImage;
 exports.buildDetailedDomSelectionPrompt = buildDetailedDomSelectionPrompt;